The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git commands via Node.js scripts to manage member worktrees and integrate branches. Potential command injection is mitigated by using spawnSync with argument arrays and validating all inputs (like member IDs and branch names) against strict alphanumeric regex patterns (MEMBER_ID_SEGMENT, SESSION_ID_PATTERN, and TASK_NAME_PATTERN).
[REMOTE_CODE_EXECUTION]: The agent runs bundled Node.js scripts to manage team state. These scripts are self-contained and use only Node.js built-in modules, eliminating risks associated with untrusted third-party dependencies or external code fetching.
[DATA_EXFILTRATION]: All team state and artifacts are stored locally in the .omo/teams/ directory. The skill does not perform network operations or access sensitive credentials, maintaining a strictly local execution profile.
[SAFE]: The skill implements defensive programming for file system operations, including explicit checks to ensure path components are not symbolic links. This prevents directory traversal attacks and ensures the agent cannot be tricked into writing or reading files outside of the authorized workspace.

teammode